Bookmarks

Contribute

AgentScan is an open project, and anything helps: reporting an automation, improving the copy or the UI, sharpening the analysis, or finding a bug.

Already contributing

11 people

The people who have shipped code to the repositories behind AgentScan, split up by project.

    • MatteoGabrieleMatteoGabriele
    • graphierosgraphieros
    • jsorefjsoref
    • huang-julienhuang-julien
    • CorentinThCorentinTh
    • trueberrylesstrueberryless
    • cernymatejcernymatej
    • rijkvanzantenrijkvanzanten
    • MatteoGabrieleMatteoGabriele
    • trueberrylesstrueberryless
    • huang-julienhuang-julien
    • mcollinamcollina
    • bluwybluwy
    • gameromangameroman
    • MatteoGabrieleMatteoGabriele
    • huang-julienhuang-julien
  • Report an automated account

    Found an account you believe is automated? Open an issue with the username, your reasoning and any supporting evidence. Reviewers then vote on it, and once enough of them approve, the account joins the verified list with a link back to the original issue.

  • Dispute a claim

    Listings come from people: someone reports an account and reviewers vote to add or reject it, so a call can be wrong. If your account was added in error, comment on the issue linked from your profile page and explain why. Entries are removed when the case warrants it.

  • Run AgentScan on your repository

    Install the GitHub App or add the action to a workflow and the same analysis runs against your own pull requests. Repositories running it publicly show up on the Used by page.

  • Improve the detection signals

    Detection is plain event analysis, no AI involved. Every signal has a label, a point value and a reason. New signals and better thresholds are welcome: open a pull request explaining what it detects and why it points at automation.

  • Work on the ecosystem

    AgentScan is spread across multiple repositories. The website and the GitHub App live in agentscan, the GitHub Action lives in agentscan-action, the Ecosystem Activity data lives in agentscan-logs, and the detection itself lives in unveil-project/identity. Bug fixes, copy improvements and ideas are all welcome.

Not sure where to start? Open an issue or a discussion, because an idea is a contribution too.